Measuring irregularities and surface defects from printed patterns

Quality inspection and assurance is a veryimportant step when today's products are sold to markets. As products are produced in vast quantities, the interest to automate quality inspection tasks has increased correspondingly. Quality inspection tasks usuallyrequire the detection of deficiencies, defined as irregularities in this thesis. Objects containing regular patterns appear quite frequently on certain industries and science, e.g. half-tone raster patterns in the printing industry, crystal lattice structures in solid state physics and solder joints and components in the electronics industry. In this thesis, the problem of regular patterns and irregularities is described in analytical form and three different detection methods are proposed. All the methods are based on characteristics of Fourier transform to represent regular information compactly. Fourier transform enables the separation of regular and irregular parts of an image but the three methods presented are shown to differ in generality and computational complexity. Need to detect fine and sparse details is common in quality inspection tasks, e.g., locating smallfractures in components in the electronics industry or detecting tearing from paper samples in the printing industry. In this thesis, a general definition of such details is given by defining sufficient statistical properties in the histogram domain. The analytical definition allowsa quantitative comparison of methods designed for detail detection. Based on the definition, the utilisation of existing thresholding methodsis shown to be well motivated. Comparison of thresholding methods shows that minimum error thresholding outperforms other standard methods. The results are successfully applied to a paper printability and runnability inspection setup. Missing dots from a repeating raster pattern are detected from Heliotest strips and small surface defects from IGT picking papers.

More than Words? An Analysis on the Language and Structure of General Election Manifestos of the British Labour Party 1983-2005

The topic of this study is the language of the educational policies of the British Labour party in the General Election manifestos between the years 1983-2005. The twenty-year period studied has been a period of significant changes in world politics, and in British politics, especially for the Labour party. The emergence educational policy as a vote-winner of the manifestos of the nineties has been noteworthy. The aim of the thesis is two-fold: to look at the structure of the political manifesto as an example of genre writing and to analyze the content utilizing the approach of critical discourse analysis. Furthermore, the aim of this study is not to pinpoint policy positions but to look at what is the image that the Labour Party creates of itself through these manifestos. The analysis of the content is done by a method of close-reading. Based on the findings, the methodology for the analysis of the content was created. This study utilized methodological triangulation which means that the material is analyzed from several methodological aspects. The aspects used in this study are ones of lexical features (collocation, coordination, euphemisms, metaphors and naming), grammatical features (thematic roles, tense, aspect, voice and modal auxiliaries) and rhetoric (Burke, Toulmin and Perelman). From the analysis of the content a generic description is built. By looking at the lexical, grammatical and rhetorical features a clear change in language of the Labour Party can be detected. This change is foreshadowed already in the 1992 manifesto but culminates in the 1997 manifesto which would lead Labour to a landslide victory in the General Election. During this twenty-year period Labour has moved away from the old commitments and into the new sphere of “something for everybody”. The pervasiveness of promotional language and market inspired vocabulary into the sphere of manifesto writing is clear. The use of the metaphors seemed to be the tool for the creation of the image of the party represented through the manifestos. A limited generic description can be constructed from the findings based on the content and structure of the manifestos: especially more generic findings such as the use of the exclusive we, the lack of certain anatomical parts of argument structure, the use of the future tense and the present progressive aspect can shed light to the description of the genre of manifesto writing. While this study is only the beginning, it proves that the combination of looking at the lexical, grammatical and rhetorical features in the study of manifestos is a promising one.

Kernel-Based Ranking. Methods for Learning and Performance Estimation

Machine learning provides tools for automated construction of predictive models in data intensive areas of engineering and science. The family of regularized kernel methods have in the recent years become one of the mainstream approaches to machine learning, due to a number of advantages the methods share. The approach provides theoretically well-founded solutions to the problems of under- and overfitting, allows learning from structured data, and has been empirically demonstrated to yield high predictive performance on a wide range of application domains. Historically, the problems of classification and regression have gained the majority of attention in the field. In this thesis we focus on another type of learning problem, that of learning to rank. In learning to rank, the aim is from a set of past observations to learn a ranking function that can order new objects according to how well they match some underlying criterion of goodness. As an important special case of the setting, we can recover the bipartite ranking problem, corresponding to maximizing the area under the ROC curve (AUC) in binary classification. Ranking applications appear in a large variety of settings, examples encountered in this thesis include document retrieval in web search, recommender systems, information extraction and automated parsing of natural language. We consider the pairwise approach to learning to rank, where ranking models are learned by minimizing the expected probability of ranking any two randomly drawn test examples incorrectly. The development of computationally efficient kernel methods, based on this approach, has in the past proven to be challenging. Moreover, it is not clear what techniques for estimating the predictive performance of learned models are the most reliable in the ranking setting, and how the techniques can be implemented efficiently. The contributions of this thesis are as follows. First, we develop RankRLS, a computationally efficient kernel method for learning to rank, that is based on minimizing a regularized pairwise least-squares loss. In addition to training methods, we introduce a variety of algorithms for tasks such as model selection, multi-output learning, and cross-validation, based on computational shortcuts from matrix algebra. Second, we improve the fastest known training method for the linear version of the RankSVM algorithm, which is one of the most well established methods for learning to rank. Third, we study the combination of the empirical kernel map and reduced set approximation, which allows the large-scale training of kernel machines using linear solvers, and propose computationally efficient solutions to cross-validation when using the approach. Next, we explore the problem of reliable cross-validation when using AUC as a performance criterion, through an extensive simulation study. We demonstrate that the proposed leave-pair-out cross-validation approach leads to more reliable performance estimation than commonly used alternative approaches. Finally, we present a case study on applying machine learning to information extraction from biomedical literature, which combines several of the approaches considered in the thesis. The thesis is divided into two parts. Part I provides the background for the research work and summarizes the most central results, Part II consists of the five original research articles that are the main contribution of this thesis.

Open Environmental Information Upon Disclosure Request in China: The Paradox of Legal Mobilization

Taking a realist view that law is one form of politics, this dissertation studies the roles of citizens and organizations in mobilizing the law to request government agencies to disclose environmental information in China, and during this process, how the socio-legal field interacts with the political-legal sphere, and what changes have been brought about during their interactions. This work takes a socio-legal approach and applies methodologies of social science and legal analysis. It aims to understand the paradox of why and how citizens and entities have been invoking the law to access environmental information despite the fact that various obstacles exist and the effectiveness of the new mechanism of environmental information disclosure still remains low. The study is largely based on the 28 cases and eight surveys of environmental information disclosure requests collected by the author. The cases and surveys analysed in this dissertation all occurred between May 2008, when the OGI Regulations and the OEI Measures came into effect, and August 2012 when the case collection was completed. The findings of this study have shown that by invoking the rules of law made by the authorities to demand government agencies disclosing environmental information, the public, including citizens, organizations, law firms, and the media, have strategically created a repercussive pressure upon the authorities to act according to the law. While it is a top-down process that has established the mechanism of open government information in China, it is indeed the bottom-up activism of the public that makes it work. Citizens and organizations’ use of legal tactics to push government agencies to disclose environmental information have formed not only an end of accessing the information but more a means of making government agencies accountable to their legal obligations. Law has thus played a pivotal role in enabling citizen participation in the political process. Against the current situation in China that political campaigns, or politicization, from general election to collective actions, especially contentious actions, are still restrained or even repressed by the government, legal mobilization, or judicialization, that citizens and organizations use legal tactics to demand their rights and push government agencies to enforce the law, become de facto an alternative of political participation. During this process, legal actions have helped to strengthen the civil society, make government agencies act according to law, push back the political boundaries, and induce changes in the relationship between the state and the public. In the field of environmental information disclosure, citizens and organizations have formed a bottom-up social activism, though limited in scope, using the language of law, creating progressive social, legal and political changes. This study emphasizes that it is partial and incomplete to understand China’s transition only from the top-down policy-making and government administration; it is also important to observe it from the bottom-up perspective that in a realistic view law can be part of politics and legal mobilization, even when utterly apolitical, can help to achieve political aims as well. This study of legal mobilization in the field of environmental information disclosure also helps us to better understand the function of law: law is not only a tool for the authorities to regulate and control, but inevitably also a weapon for the public to demand government agencies to work towards their obligations stipulated by the laws issued by themselves.